We present the stages of query compilation from a top-to-bottom perspective. Throughout this process, all available compute units may be utilized, including the classical CPU, GPU, and QPU.

Query Parser

Query compilation involves translating a high-level query, expressed in a declarative query language (e.g., SQL), into a sequence of lower-level operations that a database management system can execute efficiently. The initial stage, handled by the Query Parser, converts the original query into a set of equivalent logical query plans. This step is crucial for capturing the semantic interpretation of the query and enabling further optimization. In our architecture, this stage can be executed on a classical CPU.

Query Optimizer (Cost Estimator)

Moving to the next stage, query optimization is critical for improving logical query plans. Leveraging a cost estimator, the Query Optimizer evaluates the different logical plans and selects the most cost-effective one. This evaluation takes into account factors such as data statistics, available indexes, and expected processing cost. Traditionally, this optimization process is performed on conventional CPUs. In this proposal, we aim to explore the feasibility of offloading computational sub-tasks to a QPU. A significant challenge in this exploration is the potential scenario where the QPU is accessed over a network, introducing higher communication latency. The resulting physical plan is then executed, requiring access not only to data but also to indexes. This step, traditionally carried out on classical processors, will be examined in this proposal for potential offloading to GPUs and QPUs.

QC-aware Indexes

A notable issue is the emerging approach of integrating specialized indexes that leverage the capabilities of quantum computing (QC-aware Indexes). Quantum computing introduces unique capabilities, and adapting traditional indexing structures to exploit quantum principles could further enhance the overall efficiency of database operations.

A major challenge in this effort is extending the cost estimator to account for execution on GPUs and QPUs. Traditionally focused on I/O operations and processing cost, the existing Cost Model relies on data statistics and available indexes. However, with the utilization of compute units beyond CPUs, a comprehensive revision of the cost model becomes essential for accurately reflecting these metrics. By exploring the offloading of critical tasks to QPUs and GPUs, combined with the integration of quantum computing-aware indexes, we aim to unlock new dimensions of efficiency and performance.

Access Interface

To further give access to quantum computing and thereby advance research in both quantum computing and data management, we provide a programming interface (API) that enables non-experts to explore and utilize the benefits of quantum mechanics. Additionally, we offer an intuitive and interactive graphical user interface (GUI) as an interface for end users.